J Comp Physiol A (2016) 202:267–276 DOI 10.1007/s00359-016-1069-0

ORIGINAL PAPER

Landing on branches in the Trachycephalus resinifictrix (Anura: )

Nienke N. Bijma1 · Stanislav N. Gorb1 · Thomas Kleinteich1

Received: 6 November 2015 / Revised: 18 December 2015 / Accepted: 22 December 2015 / Published online: 23 January 2016 © The Author(s) 2016. This article is published with open access at Springerlink.com

Abstract (Lissamphibia: Anura) are famous for Introduction their saltatory or hopping locomotion, which is related to numerous anatomical specialisations that are characteristic Frogs (Anura) are well known for their saltatory locomo- for the group. However, while the biomechanics of take-off tion, which is one of the key characteristics of the group. in frogs have been studied in detail, much less is known on Anatomical specialisations for jumping have even been how frogs land after a jump. Besides terrestrial and aquatic identified in the representative of the stem-group anuran species, several lineages of frogs adopted an arboreal life- Prosalirus bitis that dates back to the early Jurassic (Shu- style and especially the biomechanics of landing on chal- bin and Jenkins 1995; Jenkins and Shubin 1998). The evo- lenging, small, and unpredictable substrates, such as leaves lution of frog jumping has received notable attention in or branches, are virtually unknown. Here we studied the the past decades (Gans and Parsons 1966; Emerson 1978; landing kinematics of the arboreal frog Trachycephalus Zug 1985; Kargo et al. 2002; Prikryl et al. 2009; Reilly and resinifictrix (Hylidae) on a wooden stick that was used to Jorgensen 2011; Astley and Roberts 2012; Jorgensen and mimic a small tree branch. We observed two different land- Reilly 2013; Astley et al. 2013; Astley and Roberts 2014) ing behaviours: (1) landing on the abdomen and (2) attach- and provides a popular textbook example in vertebrate ment with the toes of either the forelimb or the hindlimb. In comparative biomechanics (Vogel 2003). the latter case, the frogs performed a cartwheel around the Most of these studies, however, focus on the take-off stick, while they were only attached by their adhesive toe phase and the associated anatomical specialisations of the pads. We estimated the forces that act on the toes during pelvis and hindlimbs. Much less is known on how frogs this behaviour to be up to fourteen times the body weight actually land after a jump. Generally, there seem to be two of the . This behaviour demonstrates the remark- mechanisms for landing (Essner et al. 2010): (1) touch- able adhesive capabilities of the toe pads and the body con- down with the body while the limbs are stretched out and trol of the frogs. (2) a controlled touchdown with the forelimbs. The latter received particular attention and it has been shown that Keywords Arboreal locomotion · Kinematics · frogs that land on their forelimbs anticipate the impact and Adhesion · Jumping · Biomechanics activate their forelimb muscles before touchdown (Gillis et al. 2010, 2014). Nauwelaerts and Aerts (2006) measured the forces during landing and discussed the orientation of Electronic supplementary material The online version of this the forelimb to dampen the energy of jumps at touchdown. article (doi:10.1007/s00359-016-1069-0) contains supplementary material, which is available to authorized users. Azizi et al. (2014) further demonstrated that frogs that land on their forelimbs flex their hindlimbs towards the body * Thomas Kleinteich before landing to align their centre of mass with the ori- [email protected]‑kiel.de entation of the forelimbs. Limb flexion during the jump 1 Functional Morphology and Biomechanics, Zoological in the toad Bufo marinus was suggested to be facilitated Institute, Kiel University, Am Botanischen Garten 9, by an elastic recoil mechanism (Schnyer et al. 2014). In 24118 Kiel, Germany

1 3 268 J Comp Physiol A (2016) 202:267–276 another recent study, the anatomy of the pectoral girdle was described in relation to its three-dimensional movements during landing in a toad (Griep et al. 2013). In all of these studies, landing was observed on a planar surface. Frogs, however, can be found in numerous terrestrial and aquatic habitats (Duellman and Trueb 1994) and while the more basal frog lineages are mostly living on the ground, several groups independently evolved an arboreal lifestyle (Frost et al. 2006; Wells 2010; Reilly and Jorgensen 2011). Surprisingly, besides a recent study on arboreal frogs walk- ing on thin branches (Herrel et al. 2013), patterns of arbo- real locomotion are virtually unknown. Especially the land- ing behaviour of frogs in an arboreal habitat is expected to be different from that of frogs living on the ground because arboreal species land on narrow and often unpredictable substrates, such as thin branches or leaves. Safe landing Fig. 1 Trachycephalus resinifictrix sitting on a narrow branch, adher- seems especially crucial for arboreal frogs, as missing the ing with its well-developed adhesive toe pads target can have much more severe consequences than when jumping on the ground. At least, climbing back up to the canopy after a missed jump will be costly for the animals in range of forces that act on the adhesive toe pads during terms of energy consumption. It seems likely that arboreal landing in vivo. frogs show a specialised landing behaviour on narrow sub- strates that is not exhibited by terrestrial species landing on planar surfaces. Materials and methods Several lineages of frogs have adhesive toe pads that evolved multiple times within the Anura but show a remark- Specimens ably high structural similarity among different groups of frogs (Noble and Jaeckle 1928; Emerson and Diehl 1980; Four specimens of Trachycephalus resinifictrix (Goeldi Barnes et al. 2013; Drotlef et al. 2015). While these toe 1907) were available for this experiment. This species pads (or parts of them) can also be present in ground- of frog is often kept as a pet and traded under the name dwelling species (Noble and Jaeckle 1928; Manzano et al. Amazon milk frog, but it is also commonly referred to 2007), they are generally considered to be an adaptation to as Mission golden-eyed treefrog (Frank and Ramus climbing and consequently climbing is hypothesised to be 1995; Frost 2015). The specimens were captive bred the main reason why the pads are well developed in arbo- individuals that we purchased from the local pet trade. real frogs (Barnes 2007). However, besides climbing, these The animals ranged from 5.1 to 6.4 cm in snout–vent toe pads are likely to play an important role during arboreal length (SVL) and weighed between 7 and 19 g. Because landing because they can be used to produce a safe grip and the body weight in the animals was highly depend- strong damping for these animals at touchdown. The con- ent on food intake and the fullness of the bladder, we tribution of toe pad adhesion during landing has never been weighed the frogs for each experimental trial (Supple- shown before. mentary Table 1). The animals have a bluish coloration Here we used high-speed videography and kinematic with a pattern of beige or brown stripes on their back analysis to observe the landing behaviour of Trachycepha- (Fig. 1), which were used to identify individuals. Two lus resinifictrix (Anura: Hylidae) on a thin substrate. T. res- of the animals were identified as mature males with inifictrix is an arboreal species that is native to South Amer- well-developed vocal sacs and nuptial pads. The two ica (AmphibiaWeb 2015). These frogs with well-developed other individuals were females, because based on their toe pads (Fig. 1) originally only occur in primary forests, age they should have reached sexual maturity, but they where they live high up in the canopy (Hödl 1991). The clearly lacked any male characters. We kept the frogs in aims of this study were (1) to describe and quantify the a 50 50 100 cm (width depth height) terrarium × × × × movements of T. resinifictrix during landing on a narrow at a relative humidity of 70–90 %, an ambient tempera- stick, (2) to estimate the velocities of the frogs before land- ture of 26–29 °C, and with a 12-h daylight period. The ing, and (3) to estimate the effectiveness of adhesive pads animals were fed twice a week ad libitum with crickets under the typical behavioural situation and at the natural (Gryllus bimaculatus).

1 3 J Comp Physiol A (2016) 202:267–276 269

Experimental setup the wooden stick at an angle of 20°. To make the frogs jump, we placed them by hand on the lab jack and gently We performed two sets of experiments for which we tapped their hind legs. For each individual, we recorded ten recorded the movements of the frogs with a high-speed experimental trials. video camera (Photron Fastcam 1024 PCI, Photron Europe Ltd., West Wycombe, Bucks, UK) at 1000 frames per sec- Data analysis ond: (1) free hanging, and (2) landing after a jump. For scaling, we further recorded a ruler that was placed at the To analyse the high-speed video recordings, we used the same position as the frogs during the experiment with the landmark tracking plug-in MTrackJ (Meijering et al. 2012) identical camera settings as during experimental trials. In for the image analysis software Image J 1.48 (available at both experiments, we used a wooden cylindrical stick with http://www.imagej.nih.gov/ij/). For each experiment, we a diameter of 1.0 cm as target surface for the frogs. By adjusted the pixel to cm ratio based on the video record- using a white-light interferometer (New View 6000, Zygo ings of the ruler. Each landing video sequence was then Corporation, Middlefield, CT, USA) we determined the analysed frame by frame, which at a recording frequency root mean square roughness of this stick to vary between of 1000 frames per second resulted in a temporal resolution 1.3 and 4.3 µm. of 1 ms. For the experimental trials in which the frogs were free hanging at the stick, we tracked the position of the toe Free hanging pads attached to the stick, to identify sliding movements. However, it turned out that in all experimental trials, the toe We placed the frogs by hand underneath a horizontally ori- pads remained stationary once they are attached and sliding ented wooden stick and allowed them to place one hand on was never observed. For landing trials we traced the posi- the side face of the stick. We then removed our hands and tion of the rostral tip of the nasal capsule and the position of let the frogs hang free from the stick. Immediately after we the toes in lateral view. MTrackJ outputs the x- and y-coor- removed the hand, the frogs pulled themselves up towards dinates of each track of landmarks, as well as the travelled the stick, while we captured the movements of the frog in distances, and velocities over time. The velocity data in dorsal view from a perspective perpendicular to the stick. MTrackJ corresponds to the travelled distance per frame, i.e. per ms. For further analysis, we imported these data Landing into the statistical computation software R 3.1.1 (available at http://www.r-project.org). For each trial we calculated Figure 2 shows the experimental setup used to capture the a linear regression of distance over time for times rang- landing movements of the frogs. We launched the frogs ing from the onset of the recorded videos until the frogs from a Swiss Boy lab jack (Grauer AG, Degersheim, Swit- came in contact with the target. The slope of this regres- zerland), which we adjusted to a height of 35 cm. In a dis- sion is a measure of the average velocity of the frogs during tance of 25 cm, we mounted a wooden stick horizontally in the approach, which is different from the frame-by-frame the same height as the lab jack by using a laboratory stand. velocities that MTrackJ calculated (Fig. 3a; for regression The wooden stick was facing towards the high-speed video statistics see Supplementary Table 1). Further, for trials in camera with which we recorded the approach by the frogs which the frogs were attached to the wooden stick with in lateral view. To identify grip types which the frogs used their forelimbs, we further calculated a linear regression of for landing, we placed a mirror (size: 20 20 cm) above changes in velocity over time for the deceleration phase, ×

Fig. 2 Schematic drawing of the experimental setup for land- ing trials. A wooden cylindrical stick was placed 25 cm away from a platform that we used to launch the frogs. Platform and target were at a height of 35 cm. A mirror (20 20 cm) was mounted with× an inclination of 20° to the front to allow for a dorsal view. Using a high-speed video camera, we filmed the frogs while landing on the wooden stick in lateral view

1 3 270 J Comp Physiol A (2016) 202:267–276

Fig. 3 Regression analysis to calculate mean velocities during the of the frog is calculated based on the flight path (distance over time) approach (a) and negative accelerations after contact (b). The red until the first contact with the stick is reached (lower arrow). The lines show the regression function; the green line in b shows a locally frog maintains most of its forward speed until the arm is completely weighted polynomial regression of the raw data (lowess function in R stretched (distance between both arrows) and then starts to deceler- stats package), which is used for visual inspection of the graph only; ate. b The deceleration during landing is calculated from the moment calculations are based on the raw data. In the example shown here of the first attachment with the stick, until the frog gets redirected on (frog ID2, trial 1), the frog leaped over the target and reached back- a circular path. After slowing down to its minimum velocity, the frog wards with its forelimb to make contact with the stick. a The velocity accelerates and decelerates in an oscillating movement i.e. after the frogs made contact with the stick but before and thus this measurement had to be excluded from the cal- oscillation movements occurred (Fig. 3b; Supplementary culation of the average free hanging time. Table 2). The slope of this regression is a measure for the deceleration the frogs experienced, which multiplied by the Landing mass of the animals gives an estimation of the forces acting on the toe pads during this phase of landing. To study the kinematics of landing on a thin substrate, we had T. resinifictrix jump onto a horizontally oriented wooden stick with 1 cm diameter and filmed a total of 40 Results approaches in lateral view with a high-speed camera at 1000 frames per second. We never observed frogs miss- Free hanging ing the wooden stick in our experiment nor did we see trials where the animals failed to arrest the jumps. Kin- To estimate the performance of the toe pads in T. resin- ematic analysis of the video sequences revealed that the ifictrix, we first tested whether the frogs were able to sup- frogs approached the stick with an average velocity of port their body weight if hanging on horizontally oriented 1.34 0.19 m/s (N 40; Supplementary Table 1). We ± = cylindrically shaped branches using their individual toes. found that the frog ID1 was significantly slower during the We found that all specimens tested (N 4) were able to approach than the ID2 but we did not observe fur- = carry their own body weight on only two digits on either ther statistically significant differences between the animals forelimb, if the contact was made by digits three and four. (one-way ANOVA in combination with Tukey’s honest sig- In one case, the frog used only one toe pad (digit three) nificant difference test; F 3.38, df1 3, df2 36, level = = = for free hanging. Once attached to the stick, the toes were of significance p < 0.05). While the frogs are in the air, not moving or slipping down. However, usually the frogs their limbs are widespread to increase the reach between started to pull themselves up or brought the second fore- their arms and legs, respectively, and also to stabilise their limb into contact immediately after the onset of the experi- flight. ment (Supplementary Video 1). Thus, the time of free After the aerial phase, we observed two different landing hanging was limited and ranged from 51 to 277 ms (N 7; strategies (Fig. 4): (1) landing on the abdomen (i.e. a belly- = average free hanging time: 141 90 ms). In one case, the flop) (N 20; Fig. 4a; Supplementary Video 2), or (2) land- ± = frog hung for approximately 5 s, which exceeded the time ing by using the adhesive toe pads on the limbs (N 20; = we were able to record with our experimental setup (1.54 s) Fig. 2b–d). Either the toe pads of the forelimb (N 16; = 1 3 J Comp Physiol A (2016) 202:267–276 271

Fig. 4 Representative trajectories for the two different landing strate- for landing. If the frogs land by using their limbs, they differ by a gies on narrow substrates. Jumps are from right to left. Landing on variation of the position of the body relative to the target: b leaping the abdomen (a) and landing by using their adhesive toe pads on their over the target, reaching towards the stick backwards. c, d Jumping limbs (b–d). The frogs can use their forelimbs (b, c) or hindlimbs (d) too short and reaching the target by moving the limb forwards

Fig. 4b, c; Supplementary Video 3) or the hindlimb (N 4; In experimental trials in which the frogs touched the tar- = Fig. 4d; Supplementary Video 4) were used for attachment get stick with one of their limbs to initiate landing, the toe to the stick. Further, we found variation in the position pads in contact did immediately stick to the target without of the frogs relative to the target during landing with the sliding along the substrate. During these trials, the frogs limbs. The frogs either leaped over the target and reached first proceeded on their parabolic jump trajectory after one backwards with the trailing forelimb (N 8; Fig. 4b; Sup- limb was in contact with the target until this limb was fully = plementary Video 3) or descended before they reached the stretched. Then, due to the fixation of the animal at the toe target in which case they reached forwards with the lead- pads in contact, the frogs were redirected on a circular path ing forelimb (N 8; Fig. 4c; Supplementary Video 5) or and performed a cartwheel movement (Fig. 4b–d; Supple- = hindlimb (N 4; Fig. 4d; Supplementary Video 4). mentary Videos 3–5). In some cases, this cartwheel allowed = If the frogs used the limbs for attachment, they always the frogs to land on top of the stick; in most cases, however, performed a yaw movement before touchdown to orient the frogs started to oscillate like a pendulum mounted to their limbs towards the target; in some cases this behaviour the stick. During this oscillating motion, the frogs pulled was also observed during abdominal landing and helped themselves up towards the stick. Generally after one or two the frogs to orient themselves parallel to the target. Alterna- swings, the frogs brought a second limb into contact with tively, if landing on the abdomen, the frogs did also touch- the target to manoeuvre themselves on top of the stick. down in perpendicular orientation to the target in which We identified eight different grip types on the fore- case the body folded around the stick, which decelerated limb, which the frogs used to cling themselves to the target the frog immediately (Fig. 4a). After impact, the frogs while landing (Table 1). In total, we observed 27 landings immediately grabbed the stick with their adhesive toe pads in which the forelimbs were used for deceleration, either to hold on to the target. by direct contact or after the frogs touched down on the

1 3 272 J Comp Physiol A (2016) 202:267–276

Table 1 Grip types observed during landings of T. resinifictrix on a wooden stick Landings for which toes were Frequency of grip type (D digit; *D5 only at hindlimb) used (N 31) = = D2 D4 D12 D23 D24 D34 D123 D234 D5* D2345*

Abdomen (N 11) 5 (45 %) 2 (18 %) 2 (18 %) 2 (18 %) = Forelimb Reaching forward (N 8) 1 (12.5 %) 1 (12.5 %) 2 (25 %) 1 (12.5 %) 1 (12.5 %) 2 (25 %) = Reaching backward (N 8) 2 (25 %) 1 (12.5 %) 1 (12.5 %) 4 (50%) = Hindlimb Reaching forward (N 4) 1 (25 %) 3 (75 %) =

Table 2 Calculated negative accelerations and forces experienced by T. resinifictrix during attachment with the forelimb Specimen ID (trial #) Mass of specimen (g) Acceleration (m/s2) Acceleration per gravity acceleration Force (N) Force (% body weight)

2 (1) 17 19.31 2.0 0.328 196.7 − − − 2 (2) 17 13.72 1.4 0.233 139.7 − − − 2 (4) 17 77.00 7.8 1.309 784.9 − − − 2 (5) 17 20.55 2.1 0.349 209.3 − − − 2 (10) 18 6.10 0.6 0.110 62.3 − − − 3 (5) 9 141.56 14.4 1.274 1443.0 − − − 3 (6) 7 109.64 11.2 0.767 1116.9 − − − 4 (1) 13.5 52.02 5.3 0.702 530.1 − − − 4 (2) 11 28.35 2.9 0.312 289.1 − − − 4 (3) 11 52.57 5.4 0.578 535.6 − − − 4 (5) 14 33.08 3.4 0.463 337.1 − − − 4 (7) 14 13.38 1.4 0.187 136.2 − − − abdomen; in nine cases, the frogs were stopped by their corresponds to up to 14.4 times the gravitational accelera- abdomen without notable attachment of the forelimbs. tion (calculated from 141.6/9.81 14.4). Calculation of = We found that generally all digits on the forelimb can be deceleration values times the mass of the frogs gives an used for attachment. However, from the 27 landings on estimate of the forces that act on the toe pads during land- which the forelimb was involved (including cases where ing. We found that toe pads withstand forces between 0.11 the abdomen touches down first), in 13, i.e. almost 50 %, and 1.27 N (on average 0.55 0.40 N, N 12, Table 2), ± = of the observed trials only digits D3 and/or D4 were used which corresponds to 62 % and up to 1443 % of the body (Table 1). In cases in which the hindlimb first held onto the weight of the animals (Table 2). target, we found two different grip types involving digit D5 (N 1) or digits two, three, four, and five simultaneously = (N 3) (Table 1). Discussion = To estimate the forces that act on the toes of the animals when the frogs are only attached by one of their fore- or For the arboreal frog T. resinifictrix, safe landing is essen- hindlimbs during landing, we calculated the deceleration tial since they are living in heights of more than 10 m (Hödl of the frogs by linear regression of velocity over time. In 1991; Honigs et al. 2014). A failure in landing could be eight cases the statistical support for the linear regression fatal or at least a loss of height would increase the stresses was poor (p > 0.05; Supplementary Table 2) and these were on the locomotor apparatus for landing (Günther et al. excluded from the analysis. In the remainder of the trials, 1991). Further, it would be energetically very costly for deceleration ranged from 6.1 m/s2 (specimen ID2 leaping these animals to climb back. The target (i.e. a branch or a over the target and reaching backwards with digit D4 of the leaf) itself might change its position while the frog is in the forelimb) to 141.6 m/s2 (specimen ID3 reaching forward air (e.g. due to wind) or at impact of the frog, which makes with the forelimb and making contact with digit D4) and it challenging for the animals to estimate their jumping tra- was on average 47.3 42.4 m/s2 (N 12, Table 2). This jectories a priori. We found that T. resinifictrix overcomes ± = 1 3 J Comp Physiol A (2016) 202:267–276 273 these challenges by showing a high plasticity on the choice on the toe pads are beyond the body weight of the animals, of a landing strategy. Strikingly, in cases where the limbs we never observed sliding movements of the toe pads after are used, the frogs performed a partial cartwheel around the the limbs were in contact with the surface. Thus, in vivo landing stick, which demonstrates the excellent adhesive- friction between one or two toe pads and the target surface ness of their toe pads and the body control of these animals. is suitable to hold the entire animal during deceleration. To our knowledge, a similar behaviour has never been doc- The forces during deceleration, which we estimated herein umented in frogs before. (up to fourteen times the body weight of the frog), are nota- It has been previously shown that the landing strategies bly higher than the shearing forces previously reported for of frogs on plain surfaces can differ significantly between tree frog toe pads. These forces were reported to be in the species. This is in contrast to the take-off phase that range of the body weight of the frogs in animals with a appears to be largely conserved across anurans (Emerson comparable size to T. resinifictrix (Barnes et al. 2006). One and De Jongh 1980; Zug 1985; Essner et al. 2010). More part of this variation might be explained by differences in basal frogs, like Ascaphus montanus, tend to land on their the target surface. In previous experiments, the frogs were abdomen, although depending on the angle of attack during placed on a smooth Perspex thermoplastic sheet (Barnes landing the fore or hind-limbs can touch the surface first et al. 2006) that probably has a lower frictional coefficient (Essner et al. 2010). Toads (Bufonidae) and so-called true in contact with the frogs’ toe pads than the rough wood sur- frogs (Ranidae) in contrast, land balanced and stable with face. Further, Hanna and Barnes (1991) reported that meas- their forelimbs first (Nauwelaerts and Aerts 2006; Griep ured frictional forces on toe pads increased if the tilting et al. 2013; Azizi et al. 2014; Gillis et al. 2014). Here, when platform that was used for force measurements was moved landing on a stick, we observed both landing mechanisms rapidly. So after the contact between the toe pads and the (on the abdomen and landing with the forelimbs) in T. res- target is established, the viscoelasticity of the toe pads will inifictrix. In contrast to toads, where the limbs are moved allow for stronger frictional forces at high velocities. How- close to the body before impact to ensure stable landing ever, during landing, viscoelasticity of the toe pads might (Azizi et al. 2014), T. resinifictrix always stretches its limbs actually cause resistance during contact formation. In out during the jump. This is similar to the behaviour of A. any case, the contribution of the pad viscoelasticity to the montanus, which during landing on the ground results in highly dynamic processes of jump start and landing should a belly flop (Essner et al. 2010). However, in the arboreal be considered in future experimental studies. scenario, a flat body posture with stretched out limbs will Besides differences in frictional coefficients due to the stabilise the flight and also increases the chances for the difference of experimental setups and the dynamics of the frog to make contact with a target. movement in the previous publications and the present Both strategies, landing on the abdomen versus attach- study, the frogs in our experiment seemed to further max- ing with the toe pads on the forelimbs or hindlimbs, have imise frictional forces by placing the attached toes on the advantages and disadvantages for the frogs. By landing face of the target that is opposite to their body and wrap- on the abdomen, the exact estimation of the target posi- ping their digits around the stick (Figs. 4, 5; Supplemen- tion in space seems less critical as the chances of missing tary Videos 3 and 5). The Capstan Equation demonstrates the target are minimised. Further, overshooting the target the relationship between friction in the contact area and the is less likely, as the abdomen of the frog will immediately hold force: stop the flight, while if only the toes are in contact with the F = F ∗ eµθ target, the frogs proceed on their trajectory. However, dur- 1 0 (1) ing abdominal landing, the abdomen has to dissipate all with F1 as the loading force (i.e. the body weight of the of the frog’s kinetic energy, which could potentially cause frog), F0 as the holding force (i.e. the force acting on the harm to visceral organs. If the first contact is established toe), µ as the frictional coefficient between two contacting with the adhesive toe pads, however, kinetic energy will be solids, and θ as the angle that is swept by the holding limb. dissipated during the cartwheel movement. This method Thus, wrapping the limb around the stick (i.e. enlarging demands for strong adhesion at the toe pads and for a the angle θ) has a similar effect as an increase in the fric- placement of the toes on the target. tional coefficient and allows for higher loading forces with Here we demonstrated that even a single toe pad is sticky the same amount of holding force (Jung et al. 2008). In T. enough to hold the body weight of the frogs on the curved resinifictrix, after the toes are in contact with the target, substrate. It was previously shown that shearing, respec- the frogs still continue to move on the jumping trajectory tively frictional forces (i.e. forces acting parallel to the before deceleration. During this phase, however, the limb in contact surface), are important to secure the attachment of contact already wraps around the stick, which increases the tree frog toe pads (Barnes et al. 2006, 2008; Endlein et al. effective frictional forces (Fig. 5). Based on the high-speed 2013). Even in the case of landings, where the forces acting videos we recorded herein, angle θ appears to be in the

1 3 274 J Comp Physiol A (2016) 202:267–276

trajectory during the flight phase, this yaw might be caused by an asymmetrical jump, which was previously observed in primates (Jouffroy and Gasc 1974; Günther et al. 1991). In either way, this yaw movement seems to play an important role in the choice of the preferred landing strategy. For land- ing with the limbs and the cartwheel movement thereafter, it might be beneficial if the body is oriented parallel to the tar- get stick before impact to improve the grip with the toe pads. In contrast, for abdominal landing it can be advantageous to touch down with the entire width of the abdomen. We noticed individual differences between the frogs and some of these might be related to the body weight of the ani- mals. The two larger specimens (ID1 and ID2) more often performed abdominal landings compared to the smaller animals (Supplementary Table 1). Ontogenetic and scaling effects on the choice of landing strategies in frogs were not in focus here and will require further research. Further, other than scaling, experience, training level, and individual pref- erence might all influence the choice of the landing method. Besides stretching the limbs out, T. resinifictrix further increased the chances for a successful contact by exhibiting a diversity of grip types that can be applied. In our experi- ment, we observed eight different combinations of digits that were brought into contact with the target. For compar- ison, in a previous study on arboreal locomotion in frogs, Herrel et al. (2013) described only three main grip types for walking on thin branches. During landing after a jump, a precise placement of the toes is expected to be more chal- lenging than during walking. Thus, T. resinifictrix seems Fig. 5 Scheme demonstrating the increase of friction during landing to use any digit that is in a favourable position for contact, by wrapping fingers around the target stick. The dashed outline of a with no clear preference. However, similar to the observa- frog (t0) shows the early stage of contact during landing. The contact tion made by Herrel et al. (2013), digit D1 is rarely used. angle θ between stick and frog increases (θ ) as the frog continues its 0 1 Further, Herrel et al. (2013) described that digits D3 and D4 trajectory (t0). The red arrow represents the direction of this enlarge- ment are usually used by frogs walking on horizontally oriented substrates, which might be related to our results herein where we found that grip types involving these two digits range of 180°, which corresponds to π if measured in radi- are used most frequently for landing on a horizontal target. ans. The effective frictional coefficient between the limbs In summary, we showed that T. resinifictrix employs two in contact and the target stick will thus be increased more different strategies for successful landings on challenging than three times due to wrapping of the fingers around the surfaces. The choice between these two landing strategies stick. We also observed a similar behaviour during the free appears to happen during the aerial phase and is influ- hanging experiment described herein. There, the frogs also enced by the body position relative to the target. If and how increased the wrapping angle by adapting the finger to the this position is actively adjusted remains to be resolved shape of the stick. in future research projects. At contact, the toe pads of the We further found that the frogs perform a yaw movement frogs ensure an instant and strong contact to secure the of their bodies to the left or the right during the approach frogs at the target surface. The adhesion of the toe pads is before contact with the target. This yaw is always observed strong enough to allow the frogs to cling themselves to the in cases where the frogs touch down with the limbs first and target with only a few toes attached, which results in the only occasionally during abdominal landing. Because we spectacular movements we reported herein. only observed landings from a lateral view, we were not able to follow the three-dimensional trajectory of the frogs after Acknowledgments We much appreciate the insightful comments take-off and during the approach. Thus, the extent and control and the support by the Functional Morphology and Biomechan- ics group at Kiel University during this project. Discussion with of this yaw remain cryptic. Besides an active control of the

1 3 J Comp Physiol A (2016) 202:267–276 275

Alexander Filippov (Donetsk Institute for Physics and Engineering, Emerson SB, De Jongh HJ (1980) Muscle activity at the ilio-sacral National Academy of Sciences, Donetsk, Ukraine) on the geometry- articulation of frogs. J Morphol 166:129–144 controlled friction phenomena is highly appreciated. Further, we Emerson SB, Diehl D (1980) Toe pad morphology and mecha- would like to thank Stephan Bootsmann for his help with the care nisms of sticking in frogs. Biol J Linn Soc 13:199–216. of the study animals. TK is supported by a Grant of the German doi:10.1111/j.1095-8312.1980.tb00082.x Research Foundation (DFG Grant KL2707/2-1). Endlein T, Ji A, Samuel D et al (2013) Sticking like sticky tape: tree frogs use friction forces to enhance attachment on overhang- Compliance with ethical standards ing surfaces. J R Soc Interface 10:20120838. doi:10.1098/ rsif.2012.0838 All the experiments described herein have been approved by the Min- Essner RL Jr, Suffian DJ, Bishop PJ, Reilly SM (2010) Landing istry of agriculture, the environment, and rural areas of the federal state in basal frogs: evidence of saltational patterns in the evolu- Schleswig–Holstein, Germany (file number V 242-7224.121-29 (154- tion of anuran locomotion. Naturwissenschaften 97:935–939. 11/13)). All experimental trials reported herein were carried out in ac- doi:10.1007/s00114-010-0697-4 cordance with the approved guidelines. Frank N, Ramus E (1995) Complete guide to scientific and common names of and reptiles of the world. N G Publishing Conflict of interest The authors declare no competing interests. Inc., Pottsville Frost DR (2015) species of the world: an online refer- Open Access This article is distributed under the terms of the ence. Version 6.0. http://www.research.amnh.org/vz/herpetology/ Creative Commons Attribution 4.0 International License (http://crea- amphibia/. Accessed 10 Sep 2015 tivecommons.org/licenses/by/4.0/), which permits unrestricted use, Frost DR, Grant T, Faivovich J et al (2006) The amphibian tree of life. distribution, and reproduction in any medium, provided you give B Am Mus Nat Hist 297:1–370 appropriate credit to the original author(s) and the source, provide a Gans C, Parsons TS (1966) On the origin of the jumping mechanism link to the Creative Commons license, and indicate if changes were in frogs. Evolution 20:92–99 made. Gillis GB, Akella T, Gunaratne R (2010) Do toads have a jump on how far they hop? Pre-landing activity timing and intensity in forelimb muscles of hopping Bufo marinus. Biol Lett 6:486–489. doi:10.1098/rsbl.2009.1005 References Gillis G, Ekstrom L, Azizi E (2014) Biomechanics and control of land- ing in toads. Int Comp Biol 54:1136–1147. doi:10.1093/icb/icu053 AmphibiaWeb (2015) Information on amphibian biology and con- Griep S, Schilling N, Marshall P et al (2013) Pectoral girdle move- servation [web application]. http://www.amphibiaweb.org. ments and the role of the glenohumeral joint during landing in Accessed 10 Sep 2015 the toad, Rhinella marina (Linnaeus, 1758). Zoomorphology Astley HC, Roberts TJ (2012) Evidence for a vertebrate catapult: 132:325–338. doi:10.1007/s00435-013-0189-0 elastic energy storage in the plantaris tendon during frog jump- Günther MM, Ishida H, Kumakura H, Nakano Y (1991) The jump as ing. Biol Lett 8:386–389. doi:10.1098/rsbl.2011.0982 a fast mode of locomotion in arboreal and terrestrial biotopes. Astley HC, Roberts TJ (2014) The mechanics of elastic load- Zeitsch Morphol Anthropol 78(3):341–372 ing and recoil in anuran jumping. J Exp Biol 217:4372–4378. Hanna G, Barnes WJP (1991) Adhesion and detachment of the toe doi:10.1242/jeb.110296 pads of tree frogs. J Exp Biol 155:103–125 Astley HC, Abbott EM, Azizi E et al (2013) Chasing maximal perfor- Herrel A, Perrenoud M, Decamps T et al (2013) The effect of sub- mance: a cautionary tale from the celebrated jumping frogs of Cala- strate diameter and incline on locomotion in an arboreal frog. J veras County. J Exp Biol 216:3947–3953. doi:10.1242/jeb.090357 Exp Biol 216:3599–3605. doi:10.1242/jeb.090027 Azizi E, Larson NP, Abbott EM, Danos N (2014) Reduce torques and Hödl W (1991) Phrynohyas resinifictrix (Hylidae, Anura)—calling stick the landing: limb posture during landing in toads. J Exp behavior. Wiss Film (Wien) 42:63–70 Biol 217:3742–3747. doi:10.1242/jeb.108506 Honigs S, Meßing M, Pelzer B (2014) Krötenlaubfrösche – die Gat- Barnes WJP (2007) Functional morphology and design constraints of tung Trachycephalus. Natur und Tier-Verlag GmbH, Münster smooth adhesive pads. MRS Bull 32:479–485 Jenkins FA Jr, Shubin NH (1998) Prosalirus bitis and the anuran cau- Barnes WJP, Oines C, Smith JM (2006) Whole animal measure- dopelvic mechanism. J Vertebr Paleontol 18:495–510. doi:10.10 ments of shear and adhesive forces in adult tree frogs: insights 80/02724634.1998.10011077 into underlying mechanisms of adhesion obtained from studying Jorgensen ME, Reilly SM (2013) Phylogenetic patterns of skeletal the effects of size and scale. J Comp Physiol A 192:1179–1191. morphometrics and pelvic traits in relation to locomotor mode in doi:10.1007/s00359-006-0146-1 frogs. J Evol Biol 26:929–943 Barnes WJP, Pearman J, Platter J (2008) Application of peeling theory Jouffroy FK, Gasc JP (1974) A cineradiographical analysis of leaping to tree frog adhesion, a biological system with biomimetic impli- in an african prosimian (Galago alleni). In: Jenkins FA (ed) Pri- cations. Eur Acad Sci E-Newslett 1:1–2 mate locomotion. Academic Press, New York, pp 117–142 Barnes WJP, Baum M, Peisker H, Gorb SN (2013) Comparative Jung JH, Pan N, Kang TJ (2008) Capstan equation including bending Cryo-SEM and AFM studies of hylid and rhacophorid tree frog rigidity and non-linear frictional behavior. Mech Mach Theory toe pads. J Morphol 274:1384–1396. doi:10.1002/jmor.20186 43:661–675. doi:10.1016/j.mechmachtheory.2007.06.002 Drotlef DM, Appel E, Peisker H et al (2015) Morphological studies Kargo WJ, Nelson F, Rome LC (2002) Jumping in frogs: assessing of the toe pads of the rock frog, Staurois parvus (family: Rani- the design of the skeletal system by anatomically realistic mode- dae) and their relevance to the development of new biomimeti- ling and forward dynamic simulation. J Exp Biol 205:1683–1702 cally inspired reversible adhesives. Interface Focus 5:20140036. Manzano AS, Fabrezi M, Vences M (2007) Intercalary elements, doi:10.1098/rsfs.2014.0036 treefrogs, and the early differentiation of a complex system Duellman WE, Trueb L (1994) Biology of amphibians. John Hopkins in the Neobatrachia. Anat Rec (Hoboken) 290:1551–1567. University Press, Baltimore, London doi:10.1002/ar.20608 Emerson SB (1978) Allometry and jumping in frogs: helping the Meijering E, Dzyubachyk O, Smal I (2012) Methods for cell and twain to meet. Evolution 32:551–564. doi:10.2307/2407721 particle tracking. In: Conn PM (ed) Imaging and spectroscopic

1 3 276 J Comp Physiol A (2016) 202:267–276

analysis of living cells—optical and spectroscopic techniques. Schnyer A, Gallardo M, Cox S, Gillis G (2014) Indirect evidence for Elsevier, Amsterdam, pp 183–200 elastic energy playing a role in limb recovery during toad hop- Nauwelaerts S, Aerts P (2006) Take-off and landing forces in jumping ping. Biol Lett 10:20140418. doi:10.1098/rsbl.2014.0418 frogs. J Exp Biol 209:66–77. doi:10.1242/jeb.01969 Shubin NH, Jenkins FA (1995) An early jurassic jumping frog. Nature Noble GK, Jaeckle ME (1928) The digital pads of the tree frogs. A 377:49–52. doi:10.1038/377049a0 study of the phylogenesis of an adaptive structure. J Morphol Vogel S (2003) Comparative biomechanics: life’s physical world. 45:259–292. doi:10.1002/jmor.1050450109 Princeton University Press, New Jersey Prˇikryl T, Aerts P, Havelková P et al (2009) Pelvic and thigh muscula- Wells KD (2010) The ecology and behavior of amphibians. University ture in frogs (Anura) and origin of anuran jumping locomotion. J of Chicago Press, Chicago Anat 214:100–139. doi:10.1111/j.1469-7580.2008.01006.x Zug GR (1985) Anuran locomotion: fatigue and jumping perfor- Reilly SM, Jorgensen ME (2011) The evolution of jumping in frogs: mance. Herpetologica 41:188–194. doi:10.2307/3892257 morphological evidence for the basal anuran locomotor condi- tion and the radiation of locomotor systems in crown group anu- rans. J Morphol 272:149–168. doi:10.1002/jmor.10902

1 3